Table 4.

The Genetic Programming (GP) parameters used for the evolution of the binary classifiers for the jelly recognition task.

Mathematical Primitives	$\{+,-,\ast ,{/}^{\ast },{\mathit{sqrt}}^{\ast },{\mathit{log}}^{\ast },\mathit{sin},\mathit{cos},\mathit{tan},\mathit{atan}\}$
variables	the image-features summarized in Table 2
constants	k random numbers from the range $[-10,10]$,
	where k is randomly selected in the range $[0,10]$
initial population	ramped half-and-half
individual max depth	4
population size	1000
max generations	500
raw fitness	defined in the Equation (5)
scaled fitness	linear scaling
selector method	roulette wheel
crossover rate	0.9
mutation rate	$2\times {10}^{-4}$
elitism	true
termination criterion	max generations or raw fitness = 0.00